Q&A WITH EDITOR WILLIAM MEISEL William S. Meisel is the publisher and editor of Speech Recognition Update, a monthly newsletter that details developments in the industry. He has a PhD in electrical engineering and has taught at the University of Southern California in Los Angeles. He has also conducted research in speech-recognition technology. Based in Tarzana, Calif., he's also a consultant to the industry. Business Week Senior Writer Karen Pennar recently spoke with Meisel about the current state of speech-recognition technology. Here are excerpts from their conversation: Q: Is speech-recognition technology coming of age now? A: There was a lack of maturity. For years people were saying it will really come through. Then, when it was ready for prime time, no one was quite sure it was. Companies [like Microsoft] had back-burnered it. Investment in Internet technology took precedence. Meanwhile, speech-recognition technology continued to mature, but the market was slow to recognize it. Now the technology has become quite capable, and it hasn't been overhyped. Q: What about the various applications? A: Telecom applications: Each application is smart in a narrow area. Most of the programs are kind of idiot savants. While they're operating in their narrow area, they sound smart. Ask them what time it is, and they fall apart. One basic program around since 1990 or so is the application that completes collect calls. Now, almost every one of these is handled automatically. That saves millions in operator costs. There are interactive voice-response systems for call centers. There are a bunch of intermediate capability systems, such as Schwab's voice-broker system, which can respond to pretty complicated phrases. Another growth category is the voice-activated autoattendent. That should be a major product category next year. When you call a company and don't know the person's extension, and don't know how to spell their name, you can just say the name of the person you're trying to contact. The Boston Globe is using that internally. They're one of the first users. A system like that pays for itself in a year. And there are PC applications. There's excitement about the dictation systems. People aren't making jokes anymore about funny mistakes. You train the system, and tune it to your accent. Dragon Systems and IBM are neck-and-neck in this marketplace, and IBM has been very aggressive about pricing this. It's also aggressive about promoting speech interfaces, letting other people be inventive and come up with other creative uses that will help promote the use of speech recognition on PCs. Q: Will speech-recognition technology supplant the Internet, or will companies upgrade Internet capability? A: If you're an executive who happens to sit at a computer directly connected to the Internet, you have a different impression of how easy it is to get info off the Internet than most people do. For most of us, it's definitely not instantaneous, and most of us don't carry PCs around with us. If you want an instantaneous connection, and it's automated, and you don't have to wait for an attendant, then you want the telephone. And if you can make a transaction after getting info on the phone, that's going to be a lot more convenient for most people than the Internet. Over time, of course, the two technologies will merge. When you have an Internet device with a microphone, you'll be able to treat it much as you would the telephone. But the telephone is there right now, and you don't have to replicate the hardware at everybody's house. Return to main story

Updated Feb. 12, 1998 by bwwebmaster
Copyright 1998, by The McGraw-Hill Companies Inc. All rights reserved.
Terms of Use